This is a “Big” announcement.
BigQuery, one of our most popular destinations, is now available as a source connector. It also reads 1TB in under an hour, which makes it the fastest connector Airbyte has ever shipped.
This is a major improvement over the previous community connector. Not only to speed, but also to how the data is formatted and loaded into your destination. Here’s what BigQuery Source 1.0 comes with:
1 TB in under an hour on Airbyte Cloud Dates, timestamps, JSON, and nested fields arrive as real types instead of strings Interrupted syncs pick up from the last checkpoint 1 TB in under an hour Typically when extracting data from BigQuery, you have to rely on querying, which results in a heavy pagination workload. This is not ideal, and we sought an alternative approach when building the new connector.
When using Airbyte Cloud, the BigQuery source connector uses a better method. It reads through the BigQuery Storage Read API, which streams table data as compressed batches across many parallel streams. This is a significantly more efficient process, moving 480 MB/s compared to 3.7MB/s via the standard query path.
That's how we load a TB of data in your destination in under an hour.
Typed data and syncs that pick up where they left off Previously, the community connector extracted dates, timestamps, JSON, and binary values as strings, and STRUCTs and ARRAYs arrived without a schema. Users relied on downstream models to parse the data.
The new connector keeps data as it is typed in BigQuery, no parsing required.. Dates land as dates, timestamps maintain time zones, JSON lands as JSON, and nested STRUCTs and ARRAYs keep their fields. It also reads primary keys straight from your table constraints.
Critically, we improved syncing on this connector, something we know our users will appreciate.
On Airbyte Cloud, if a BigQuery source sync fails or gets cancelled, the next run resumes from the last checkpoint (within BigQuery's six-hour read session window) instead of reading the whole table again. This is a huge timesaver, especially for your largest tables.
I’m excited to see the user experience improvements from these upgrades. We’d love to hear your feedback in the Slack community.
Start moving data from BigQuery today Source BigQuery 1.0 is live on Airbyte Cloud as an opt-in upgrade until December 1, 2026. Open your BigQuery source, click Upgrade , and refresh the schema on each connection.
Because column types change, 1.0 is a breaking change. Read the migration guide before you upgrade, especially if you run incremental syncs or have downstream models that expect these columns as strings.
For full speed, you’ll need to run it on Airbyte Cloud. On Airbyte Core, you still get the rebuilt connector with typed columns and nested schemas, but it reads through the less-efficient standard query path.
New to Airbyte? Start a free trial and move your BigQuery data to any destination.
